Biosynthesis, Glycosylation, Movement through the Golgi System, and Transport to Lysosomes by an N-Linked Carbohydrate-independent Mechanism of Three Lysosomal Integral Membrane Proteins*

The biosynthesis, glycosylation, movement through the Golgi system, transport to lysosomes, and turnover of three lysosomal integral membrane proteins (LIMPs) have been studied in normal rat kidney cells using specific anti-LIMP monoclonal antibodies. Im- munoelectron microscopy studies revealed the presence of LIMPs in secondary lysosomes, Golgi cisterna, and coated and uncoated vesicles located in the trans- Golgi area. Pulse-chase experiments recorded LIMP precursors of 27 (LIMP I), 72 (LIMP 11), and 86 kDa (LIMP 111) and mature LIMPs of 35-50 (LIMP I), 74 (LIMP 11), and 90-100 kDa (LIMP 111). Time course studies on the acquisition of endoglycosidase H resistance by LIMPs indicated that all three LIMPs moved from the site of their synthesis in the endoplasmic reticulum to the medial Golgi within 30-60 min after their synthesis. All three LIMPs were fully glycosy- lated before leaving the Golgi system, the process during which LIMP I was retained in the trans side of the organelle. LIMP I reached the lysosomes with a half- time of 2 h and LIMPs I1 and I11 with half-times of 1 h after their synthesis by a mechanism that was inde- pendent of N-linked carbohydrates. LIMPs free of N-linked carbohydrates displayed that maximal conversion of N-linked high-mannose oligosaccharides to N-linked complex carbohydrates occurred in half an hour for LIMP 111, 1 h for LIMP 11, and was not yet completed after 1.5 h of chase in the case of LIMP I. N0t.e that the endo-H-resistant form of LIMP I displayed a slight increase in molecular weight between 1 and 1.5 h of chase. This processing was completely sensitive to endo F digestion and therefore can be ascribed to processing of the complex-carbohydrate chains. mature form; P, precursor form; deglycosylated form.

The biosynthesis, glycosylation, movement through the Golgi system, transport to lysosomes, and turnover of three lysosomal integral membrane proteins (LIMPs) have been studied in normal rat kidney cells using specific anti-LIMP monoclonal antibodies. Immunoelectron microscopy studies revealed the presence of LIMPs in secondary lysosomes, Golgi cisterna, and coated and uncoated vesicles located in the trans-Golgi area. Pulse-chase experiments recorded LIMP precursors of 27 (LIMP I), 72 (LIMP 11), and 86 kDa (LIMP 111) and mature LIMPs of 35-50 (LIMP I), 74 (LIMP 11), and 90-100 kDa (LIMP 111). Time course studies on the acquisition of endoglycosidase H resistance by LIMPs indicated that all three LIMPs moved from the site of their synthesis in the endoplasmic reticulum to the medial Golgi within 30-60 min after their synthesis. All three LIMPs were fully glycosylated before leaving the Golgi system, the process during which LIMP I was retained in the trans side of the organelle. LIMP I reached the lysosomes with a halftime of 2 h and LIMPs I1 and I11 with half-times of 1 h after their synthesis by a mechanism that was independent of N-linked carbohydrates. LIMPs free of Nlinked carbohydrates displayed much shorter halflives than fully glycosylated LIMPs, suggesting an important role of the sugars in protecting LIMPs against proteolytic degradation. Double immunofluorescence microscopy experiments showed that LIMP I, LIMP 11, and LIMP I11 are localized in the same lysosomes.
The biogenesis of lysosomes is a complex process that requires the synthesis and processing of soluble and membrane components, their sorting from other newly synthesized cellular elements, and finally their integration into new or pre-existing lysosomes. In recent years great progress has been made in the knowledge of the biosynthesis, processing, and mechanism of sorting of the main soluble lysosomal components, the acidic hydrolases (reviewed in Refs. 1 and 2). These proteins are co-translationally inserted into the lumen of the endoplasmic reticulum and then transported to the Golgi system together with other secretory and membrane proteins. In this organelle they acquire mannose 6-phosphate in the N-linked high-mannose oligosaccharide chains (3)(4)(5).
Recognition of the mannose 6-phosphate residues by specific * This research was supported in part by a Fundacion Juan March Grant (to J. G. B.) and a Courtesy International Grant from Washington, D. C. (to J. S. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. membrane receptors located in the cisterna of the Golgi system results in sorting and delivery of the enzymes to lysosomes (6,7). The enzymes are proteolytically processed during their transport and upon their arrival in lysosomes (8,9;revised in Ref. 2).
Much less is known about the biosynthesis, processing, mechanism of sorting, and delivery of integral membrane proteins (LIMPs)' to lysosomes. Although recently some LIMPs have been characterized using polyclonal and monoclonal antibodies (10-13), the information about those processes is still very limited. T o address some of these questions we have obtained monoclonal antibodies against LIMPs with which we have studied the molecular characteristics, biosynthesis, processing, movement from endoplasmic reticulum to lysosomes, and degradation of three of these proteins. Pulsechase experiments show that after their synthesis LIMPs acquire N-linked high-mannose and complex carbohydrates as shown by their sensitivity to endo H and endo F digestion. Study of the rates of acquisition of endo H resistance indicate that all three LIMPs move from the cis to the trans side of the Golgi system with similar velocities. LIMPs are transported from Golgi to lysosomes with different rates by a mechanism that is independent of N-linked carbohydrates. Furthermore, LIMPs display different half-lives and the carbohydrates acquired in the Golgi system protect LIMPs against proteolytic degradation in the lysosomes. The implications of these results in the biogenesis and degradation of lysosomes are discussed.

RESULTS
Development of Monoclonal Anti-LIMP Antibodies and Cellular Localization of the Antigens-Screening of hybridoma clones for producers of anti-LIMPS antibodies was performed by indirect immunofluorescence microscopy using NRK cells.
The abbreviations used are: LIMP, lysosomal integral membrane protein; BSA, bovine serum albumin; endo H, endo-P-N-acetylglucosaminidase H; endo F, endo-(3-N-acetylglucosaminidase F GIMP, Golgi integral membrane protein; dGLIMP, lysosomal integral membrane protein without N-linked carbohydrates; PAGE, polyacrylamide gel electrophoresis; PMSF, phenylmethylsulfonyl fluoride; SDS, sodium dodecyl sulfate; NRK, normal rat kidney. Portions of this paper (including "Experimental Procedures," part of " Results," and Figs. 2,3,and 5) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 86M-2233, cite the authors, and include a check or money order for $6.00 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.  The selected clones secreted antibodies that exclusively found to stain mainly large secondary lysosomes ( Fig. 1, a and stained cytoplasmic vesicles predominantly clustered around b). Occasional staining of Golgi cisterna (Fig. 1, c and d), the nucleus (see Fig. 9). Three of these clones, 14E12,29G10, coated vesicles probably budding from the trans-cisterna, and and 38C7, produced antibodies that specifically recognized uncoated vesicles located in the trans-Golgi area were also three different LIMPs (see below). When studied by immuobserved (Fig. 1, c and d). The small size vesicles without coat noelectron microscopy in NRK cells, the antibodies were display characteristics of primary lysosomes as shown in Processing Transport Turnover of Lysosomal Membrane Proteins 16757 studies with pituitary cells. These cells in the absence of hypothalamic releasing factors displayed an extensive process of lysosome-dependent crinophagia involving secretory granules (34). As a result the cells exhibited large numbers of primary lysosomes. These organelles were positively stained with anti-LIMP antibodies and consisted of small vesicles spread throughout the cytoplasm that were frequently found in the vicinity of or fusing with secretory granules (Fig. 1, e and f). The fusion event probably starts the degradation of the granules and the formation of secondary lysosomes. Molecular Characterization of LIMPS and Localization of LIMP Epitopes-See Miniprint Section.
Hate of LIMPs Transport through the Golgi System-The acquisition of sialylated N-linked complex carbohydrates by LIMPs clearly indicated that these proteins traversed the entire Golgi system in a cis to trans direction. To determine the rate of this movement, the time course of acquisition of endo H resistance by the mature LIMPs was studied (Fig. 4). We observed that after half an hour of chase most of LIMP 111 displayed the same resistance to endo H exhibited by its mature form after 1.5 h of chase. A substantial part of LIMP I and LIMP I1 also displayed endo H resistance by this time.
By 1 h of chase all the LIMP I1 and 111 exhibited the partial resistance to endo H characteristic of their mature forms, whereas after 1 h and 1.5 h of chase small amounts of LIMP I still displayed a sensitivity to endo H that was not observed in the mature form. It is noteworthy that all the endo Hresistant forms of LIMPs were sensitive to neuraminidase after 1 h of chase (not shown). All these results indicated that LIMPs moved from the cis to the trans side of the Golgi with similar hut not identical rates. A careful study of the electrophoretic mobility of LIMPs revealed that whereas LIMP I1 and LIMP 111 did not experience changes in their molecular weights after acquiring the endo H resistance, LIMP I still exhibited a slight increase in it (Fig. 4). I t is important to note that endo F digests of LIMP I immunoprecipitated either immediately after acquiring endo H resistance or after complete processing displayed similar molecular weights (Fig. 2, panel A, compare lanes 3 with 9). Furthermore, metabolic labeling of LIMP I with the sialic acid precursor [6-:'H]Nacetylmannosamine resulted in the exclusive labeling of forms of the protein with apparent molecular mass higher than 45 kDa (data not shown). These results suggest that slow sialylation could be responsible for the retarded processing of LIMP I. Transport of LIMPs from the Golgi System to Lysosomes-To study the transport of LIMPs from the Golgi system to lysosomes the proteins were metabolically labeled with ['''SI methionine, chased for different periods of time, and their distribution between t.he low density Golgi elements and high density lysosomes analyzed after separating the organelles by Percoll gradients (35). Fig. 5 (see Miniprint Section) shows the separation of Golgi elements and lvsosomes bv Percoll gradients.
The transport of LIMPs from the Golgi system to lysosomes is described in Fig. 6. Operationally the transport rate is defined as the time required for a newly synthesized LIMP to move from the fractions containing the Golgi elements to the fractions containing lysosomes. The half-time of transport is calculated as the time required for 50% of the newly synthesized protein to consummate the movement. I t can be observed that half an hour after their synthesis, the precursors and mature forms of LIMPs were located in the fractions containing the Golgi elements. However, a small amount of mature LIMP I1 was already detected in lysosomes by that time. As the t,ime of chase was increased the amount of mature LIMPs associated with lysosomes became larger and a parallel decrease of their presence in the fractions containing the Golgi elements was recorded. These results indicated t.hat LIMPs were effectively transported from the Golgi syst.em to lysosomes. By 2 h of chase the transport of LIMPs I1 and I11 was virtually completed as indicated bv their almost exclusive localization in t.he lysosomal fractions. Interestingly, only 40 and 70% of LIMP I was found associated with lysosomes after 2 and 4 h of chase. The prolonged presence of LIMP I in the fract.ions containing Golgi and plasma membrane elements could result from retent,ion of the prot.ein in the Golgi system or transport of the protein to the plasma membrane. These two possibilities were examined in different experiments. In a first experiment, cells metabolically labeled with ["S]methionine for 15 min and chased with cold methionine for 0, 1, 2, 3, and 4 h were incubated with anti-LIMP I antibody for 1 h a t 4 "C and the incorporation of newlv synthesized LIMP I into the plasma membrane st,udied by immunoprecipitation (36). No T-labeled LIMP I was immunoprecipitated by the antibody. In a second experiment, cells labeled as before were incubated for 30 min a t 4 "C in the absence and presence of proteinase K and the levels of "S-labeled LIMP I in the fractions of the Percoll gradient containing Golgi elements and plasma membrane compared. No significant differences in the levels of LIMP I were found between the fractions obtained from untreated and proteinase K-treated cells. These results suggested that the prolonged presence of newly synthesized LIMP I in the light fractions of the Percoll gradient was the result of its retention in the Golgi system before being delivered to lysosomes. This agrees with the observation that glycosylation of LIMP I continued after acquiring resistance to endo H, a process which is likely to result in retention of the protein in the trans side of the Golgi system. The selectivity of the transport of LIMPs to lysosomes was examined by comparing their distribution in Golgi elements and lysosomes with that of a 130-kDa Golgi cisterna integral membrane protein (GIMP), characterized in our l a b~r a t o r y .~ As can be seen in Fig. 6, there was no detectable amount of GIMP in lysosomes after 2 h of chase. This result indicated the existence of specific mechanisms of sorting and transport that selected the membrane proteins delivered to lysosomes. LIMPs from the Golgi system to lysosomes was of considerable interest for two reasons. The conceivable involvement of N-linked oligosaccharides in this process, and the possibility that the transport of LIMPs to lysosomes could be coupled to the mannose 6-phosphate-dependent transport of lysosomal enzymes. These studies were performed with the same protocol used to study the transport of LIMPs in normal cells with the difference that 5 rg/ml tunicamycin and 10 pg/ml leupeptin were continuously present in the cell culture medium. Henceforth the N-linked oligosaccharide free forms of LIMPs will be referred to as dGLIMPs. Study of the transport of dGLIMP I revealed that 80% of the protein was already in lysosomes 2 h after its synthesis (Fig. 7). A similar rate of transport was recorded for dGLIMP 111. The rapid transport of dGLIMP I was in contrast with the considerably slower rate of movement of the fully glycosylated LIMP I (see above). This difference, the evidence that LIMP I is retained in the Golgi system and the similar rates of transport displayed by dGLIMP I and dGLIMP 111 supported the suggestion that glycosylation of LIMP I and not its movement from Golgi to lysosomes was the rate-limiting step in its transport to lysosomes. It is relevant that in contrast with dGLIMP I, apparently stable during the time required to study its transport, dGLIMP 111 was immediately degraded in lysosomes. Degradation resulted in production of two 33-and 23-kDa peptides that were immunoprecipitated by the anti-LIMP I11 antibody from the lysosomal fractions. The transport of dGLIMP I1 to lysosomes could not be determined with certainty. Although a significant decrease in the amount of dCLIMP I1 associated with Golgi elements was detected in chase experiments, a parallel increase was not observed in lysosomes. This result probably reflected the rapid degradation of dGLIMP I1 in lysosomes. Nevertheless, we cannot rule out the possibility that dGLIMP I1 is not transported to lysosomes and it is degraded somewhere else. It is noteworthy that the Golgi protein, 130-kDa GIMP, was localized in Golgi elements in cells treated with tunicamycin (not shown) indicating that the absence of carbohydrate did not affect its normal distribution and was not transported to lysosomes for degradation. Therefore, the results clearly showed that dGLIMP I and 111 were effectively transported from the Golgi system to lysosomes in tunicamycin-treated cells. This was consistent with mechanisms of sorting and transport of LIMPs to lysosomes being independent of the N-glycosylation of these proteins. Also, they indicated that transport of LIMPs from the Golgi system to lysosomes was not coupled to the mannose 6phosphate-dependent transport of soluble lysosomal enzymes.

Transport of LIMPs from the Golgi System to Lysosomes in
Half-lives of LIMPs and dGLIMPs-The turnover of fully glycosylated LIMPs and N-linked carbohydrate free LIMPs (dGLIMPs) was studied in normal and tunicamycin-treated cells, respectively, in experiments in which the pulse-labeled proteins were chased for different periods of time. These experiments showed that LIMPs displayed different halflives, the shortest of them corresponding to LIMP I with 8 h and the longest one to LIMP I1 with 20 h, with LIMP 111 exhibiting a half-life of 10 h (Fig. 8). Furthermore, the halflives of dGLIMP I, 4 h, dGLIMP I1 and dGLIMP 111, 1 h, were considerably shorter than those of the fully glycosylated LIMPs (Fig. 8). These results were consistent with the hy- pothesis that N-linked carbohydrates play a role in protecting LIMPs against proteolytic degradation. This role could be of physiological importance given the specific localization of LIMPs in lysosomes and the high concentration of proteolytic enzymes in these organelles.
Co-distribution of LIMPs in Lysosomes-Having found that LIMPs were transported to lysosomes a t different rates and displayed different half-lives, we considered the possibility that these differences could reflect their localization in different lysosomes. To study this possibility we analyzed the distribution of the three LIMPs in lysosomes by double immunofluorescence microscopy. As shown in Fig. 9 these experiments revealed a complete coincidence in the cellular distribution of LIMPs, indicating that LIMPs I, 11, and 111 were located in the same lysosomes. I I;I(;. 9 . ('0-localization of 1,IMl's in the same lysosomes. NIIK cells were studied hv douhle immunofluorescence using the monoclonalanti-LIMl'antihodies 14E12 (LIMI'I) and 29G10 (LIMP 11). In a first step, the cells were incubated with anti-LIMP I antibody, this was tagged with rhodamine-conjugated goat anti-mouse antibody and both cross-linked to LIMP I using formaldehyde. In a second step, the cells were incubated with anti-LIMP I1 antibody and then with fluorescein-conjugated goat anti-mouse antibody. Complete correlation between the distribution of the two anti-LIMP antibodies was interpreted as evidence that LIMP I and LIMP I1 were localized in the same lysosomes as the tagged rhodamine-conjugated antibody was cross-linked to the LIMP I site and could not have reacted with any LIMP I1 located in different lysosomes. A, rhodamine channel; R, fluorescein channel. Note the identical cellular distribution of LIMP I and LIMP I1 in vesicles clustered around the nucleus that were identified by electron microscopy as lysosomes. Control experiments in which the first or second antibody steps were omitted revealed no rhodamine fluorescence in the fluorescein channel and vice versa. Identical results were obtained in studies of the codistribution of LIMP I and I1 with 111.

DISCUSSION
We have developed and used monoclonal antibodies against integral membrane proteins from lysosomes to study the cellular distribution, molecular characteristics, processing, movement from the endoplasmic reticulum to lysosomes, and turnover of these proteins. The aim of these studies is to increase our knowledge of the biogenesis and biodegradation of lysosomes.
The three LIMPs here characterized contain N-linked carbohydrates. LIMP I apparently contains exclusively complextype carbohydrates, whereas LIMPs I1 and I11 contain complex as well as high-mannose oligosaccharides. Sialylatedcomplex carbohydrates have been detected in all three LIMPs. We have not found any evidence of the presence of 0-linked carbohydrate chains in LIMPs as shown by their insensitivity to 0-glycanase. However, it cannot be discarded that the small increase in the apparent molecular weight of LIMP I11 during pulse-chase experiments performed in the presence of tunicamycin could correspond to the acquisition of 0-linked carbohydrates by the protein.
Several lysosomal membrane proteins have been recently characterized using monoclonal and polyclonal antibodies; two proteins, LAMP I and LAMP 11, have been identified in mouse lysosomes (11), one, Igpl20, in rat (10) and one, gp95-105, in chicken lysosomes (12).
Three of these proteins, LAMP I, Igpl20, and gp95-105 and the rat LIMP 111 here described, display precursors and mature forms with similar apparent molecular weights and isoelectric points. Although the reported cellular localizations of these proteins are not identical (LAMP 1 (37) and gp95-105 (12) are both expressed on the plasma membrane whereas LIMP I11 is exclusively found in ly~osomes)~ it cannot be excluded that they could be the same protein. It is also not clear whether any of these proteins is kindred to the 100-kDa protein from lysosomes studied with a polyclonal antibody and immunologically related to the K',H+-ATPase pump from gastric mucose (13). No monoclonal antibody has been previously obtained against LIMP 11. This protein could be related to other of similar molecular weight that has not been characterized and was immunoprecipitated with several other membrane proteins by a polyclonal antibody prepared against lysosomal membranes (10). With respect to LIMP I there is no report of a lysosomal membrane protein with its characteristics.
The movement of LIMPs from the site of their synthesis in the endoplasmic reticulum to and through the Golgi system has been examined by studying the conversion of N-linked high-mannose carbohydrates to complex carbohydrates. Only proteins reaching the trans-Golgi have their N-linked oligosaccharide chains converted to sialylated complex carbohydrates, due to the location of P-galactosyltransferase (38) and sialyltransferase (32,33) in this part of the Golgi system. The finding that LIMPs I, 11, and 111, as well as other LIMPs previously described (10,12,39) acquire N-linked complex carbohydrates containing sialic acid, strongly indicates that they cross the entire Golgi in a cis to trans direction. It is important to note that the movement of LIMPs from the endoplasmic reticulum through the cis and to the medial-Golgi, where endo H resistance is acquired (40), is performed with similar speeds (30-60 min). Furthermore, whereas the speeds are comparable to that of other membrane (41) and soluble proteins, including some lysosomal hydrolases (42), that traverse the Golgi system, they are different from those of other proteins performing the same movement (43,44). It is noteworthy that whereas the processing of LIMP I1 and I. Sandoval, unpublished results. LIMP I11 is terminated immediately after acquiring endo H resistance, that of LIMP I slowly continued. The localization of sialyltransferase in the trans-Golgi (32, 33) and the incorporation of sialic acid to the forms of LIMP I displaying high apparent molecular weights, suggest that the slow processing of LIMP I is probably due to the late acquisition of sialic acid residues. There is considerable controversy about the site in the Golgi system from which different lysosomal components are delivered to lysosomes. Lysosomal hydrolases have been proposed to be released from cis-Golgi (45), trans-Golgi (46, 47), and even from an independent organelle called GERL (48). The facts that LIMPs cross the entire Golgi system and that LIMP I1 and LIMP I11 are transported to lysosomes immediately after reaching the trans-Golgi indicates that most likely LIMPs are delivered to lysosomes from the trans side of the Golgi system.
Study of the transport of LIMPs from the site of their synthesis in the endoplasmic reticulum to a dense lysosomal compartment reveals that two of these proteins, LIMP I1 and LIMP I11 are transported with a half-time of 1 h, whereas LIMP I is transported with a half-time of 2 h. The movement of LIMP I1 and I11 from the endoplasmic reticulum to the medial side of the Golgi system in 30-60 min and their presence in lysosomes 60 min after their synthesis, indicate that these two LIMPs are transported to lysosomes immediately after their arrival to the trans-Golgi. In contrast, the delayed presence of LIMP I in lysosomes 2 h after its synthesis and 90 min after its arrival to the trans-Golgi compartment indicates that it is probably retained in the trans-Golgi (see above).
It is important to note that the rates of transport of LIMPs to lysosomes are within the range of those reported for soluble lysosomal enzymes in different systems (42,49,reviewed in Ref. 2). The different rates with which some lysosomal hydrolases are transported from the endoplasmic reticulum to lysosomes (50) could also be due to different processing in the Golgi system. However, there are important differences between the mechanisms of delivery of LIMPs and soluble lysosomal enzymes from the Golgi system to lysosomes. LIMPs are not phosphorylated and at least two dGLIMPs, dGLIMP I and dGLIMP 111, are transported to a dense lysosomal compartment in the presence of tunicamycin, a drug that completely blocks the transport of lysosomal hydrolases from the Golgi system to lysosomes (51). These results indicate that transport of LIMPs from the Golgi system to lysosomes is not mediated by the mannose 6-phosphate residues that, contained in N-linked high-mannose oligosaccharide chains, instrument the transport of lysosomal hydrolases. They also indicate that the transport of LIMPs is not coupled to that of soluble lysosomal hydrolases and that it is independent from N-linked carbohydrates. This independence is in contrast with the role played by N-linked carbohydrates in the movement of some glycoproteins to the plasma membrane (52). The absence of 0-linked carbohydrates from the LIMPs here studied also seems to discard a role of these sugars in the transport. It is likely that the putative signals for sorting and delivery of LIMPs to lysosomes reside in the amino acid sequence, tertiary structure, or in the same type of posttranslational modification of these proteins. In contrast, there is a marked effect of carbohydrate on the stability of LIMPs as demonstrated by the observation that dGLIMPs synthesized in the presence of tunicamycin display much shorter half-lives than the fully glycosylated mature LIMPs. In fact the differences probably are larger than the ones measured by us here as the half-lives of LIMPs were measured in the absence of protease inhibitors whereas those of dGLIMPs were measured in the presence of leupeptin. Clearly, dGLIMP I11 is rapidly degraded by proteolysis upon its arrival to lysosomes and although the site of the accelerated degradation of dGLIMP I and dGLIMP I1 has not been determined, it cannot be discarded that it is in lysosomes. These results are consistent with the general observation that glycoproteins are degraded faster after deglycosylation (reviewed in Ref. 53). They also point to a fundamental role of the carbohydrate moieties in protecting LIMPs from degradation by lysosomal hydrolases as has been hypothesized previously by others (10). This may also explain why in normal cells only fully glycosylated LIMPs are delivered to lysosomes. The immunoelectron microscopy studies of the cellular distribution of LIMPs in NRK cells agree with the biochemical evidence indicating that LIMPs cross the entire Golgi system before being delivered to lysosomes from the trans side of this organelle. In these studies LIMPs are found to be present in several Golgi cisternae, coated vesicles probably budding from the trans most cisterna and in uncoated vesicles located in the trans-Golgi area. It is not clear whether LIMPs are transported from Golgi to dense lysosomes, probably secondary lysosomes, in the same or different vesicles. Our data cannot distinguish between these two possibilities. Double immunoelectron microscopy studies of these vesicles, using anti-LIMP antibodies labeled with different probes could be helpful to solve this problem.
It is important that primary cultured pituitary cells displaying extensive crinophagia of secretory granules (34) contain large numbers of LIMP-loaded vesicles similar to the uncoated vesicles found in the trans-Golgi area of NRK cells. As expected, many of these vesicles are frequently found fusing with secretory granules. These observations suggest the possibility that vesicles loaded with lysosomal components after being released from the Golgi system may directly fuse with organelles selected for degradation and, perhaps, as well, with autophagic vesicles, endosomes, multivesicular bodies, and secondary lysosomes. Formation of lysosomes outside the Golgi system opens the possibilities that different lysosomal components may leave the Golgi system independently from the same or different sites and that some of the lysosomal components may never pass through the Golgi system. Also the possible existence of different types of lysosomes should not be discarded. The finding that LIMPs exhibit different half-lives, without supporting the existence of different classes of lysosomes, argues against the existence of a lysosome that is formed and degraded as a unit.